Economic Analysis of Law and the Value of Efficiency
forthcoming in Aristides N. Hatzis ed. Economic Analysis of Law: A European Perspective
(Cheltenham, U.K.: Elgar 2005)
Bruce Chapman
Faculty of Law, University of Toronto
bruce.chapman@utoronto.ca
Economic analysis of law can usefully be divided into three types (Trebilcock
1997). Two of these are positive forms of analysis, and the third is normative. The
positive analyses purport to use economics either to describe how things are, or to predict
how things will be, but they do not seek either to assess or prescribe how things should
be. Thus, in the first form of positive analysis, positive theories of legal doctrine, the
claim is that economics offers a very useful way (perhaps even the most useful way) to
understand legal doctrines as they are, and to comprehend why the law might make the
distinctions that it does. For example, the rules of negligence liability are best understood,
the argument might go, as incentives to shape the behaviour of parties so that the costs of
accidents are either efficiently avoided or distributed (Landes and Posner 1997). But no
claim is made that these rules, understood in this way, are the best rules, or that values
other than economic efficiency might not be worthy of pursuit.
Likewise, in the second form of positive analysis, legal impact analysis, where
the gaze shifts from an understanding of legal doctrine as developed by judges and other
law makers to the prediction of how economic actors will respond to various legal rules,
the claim, again, is that these are only predictions about behaviour. The economist as
positivist makes no claim that these predictions about the impact of legal rules have any
special implications for what the law should be. Thus, for example, when the economist
predicts that individuals might take less care driving their cars if there is a legal
2
obligation to wear a seatbelt, because the wearing of a seatbelt lowers the cost of an
accident to the driver when it occurs, the point stops there (Peltzman 1975). The idea is
not to suggest, as a normative matter, that we should question the propriety of having any
such legal obligation to wear our seatbelts.
However, in the third type of economic analysis, the agenda is overtly normative.
Social values are explicitly advanced as attractive ones for society to adopt in the making
of legal rules. In economics the primary value is Kaldor-Hicks efficiency (named after the
two prominent British economists who first articulated it) or wealth maximization, a
value, we shall see, that the economist has sought to derive from the closely related
values of Pareto superiority and Pareto optimality. Thus, in normative economic analysis
of law, the claim is that legal rules should be shaped so as to promote Kaldor-Hicks
efficiency, and it is grounds for criticizing a legal rule if it ignores this consideration
(Zerbe 2001). Thus, for example, if a rule of strict liability can be shown to provide better
incentives for the efficient avoidance and distribution of accident costs than the currently
prevailing negligence rule (a claim that will likely also make use of legal impact
analysis), then, all else equal (for example, the costs of litigation), that will count as a
reason for the law to adopt the strict liability rule.
In this essay, I shall try to explain in some greater detail what is entailed by a
normative prescription for law based on Kaldor-Hicks efficiency or wealth maximization,
and offer some reasons for being cautious about being guided by it. In the course of my
analysis I will have reason to emphasize the difference between wealth and welfare, and
show how a single-minded pursuit of the former can frustrate our enhancement of the
3
latter. Indeed, I will demonstrate that the systematic pursuit of wealth maximization can
make us all worse off.
My analysis will also borrow from social choice theory and suggest an economic
interpretation of the relevance that a common social consensus or community morality
might have for the enhancement of individual welfare under legal rules. In the final
analysis, this essay will show that the value of Kaldor-Hicks efficiency provides a much
less secure basis for assessing and prescribing law than the economist thinks, and that it
needs to be heavily supplemented by a concern for a common social consensus if the law
is achieve anything of real value in terms of human welfare. I will finish by raising some
more fundamental concerns about the single-minded pursuit of welfare and argue that
there may even be some reason to doubt that we should share the economist’s allegiance
to the principle of Pareto efficiency from which the principle of Kaldor-Hicks efficiency
is derived.
I.
Kaldor-Hicks Efficiency and its Paretian Pedigree
Suppose two individuals form a market contract to their mutual satisfaction, and
that the contract has no external effects, that is, that the contract for the exchange or reallocation of certain goods does not affect anyone else in any way. If the two individuals
feel better off in virtue of the contract (and why else would they choose to make the
contract?), and no one else feels any worse off (as no one else is affected), then the
contract has brought about what economists call a Pareto superior state of affairs: at least
one person is better off than she was, and no one is worse off. We can also say that the
4
contract has increased total welfare (or satisfaction); if welfare has increased for each of
the two contractors, and not been decreased for anyone else, then the total amount of
welfare must be up overall.
Note also what we cannot say. First, we cannot say that total welfare is
maximized because of the contract; we can only say that it has increased. This could be
because there are even more contracts that could be made, which further increase total
welfare. Economists refer to the state of affairs from which no further Pareto superior reallocations can be made as a state of Pareto optimality. Thus, Pareto superiority refers to
a comparison (or a move) between two states of affairs, and Pareto optimality is an
attribute of a given state of affairs (Coleman 1988).
Second, we also cannot say that the goods or services have been transferred by
contract to someone who values them more highly (in terms of welfare) than the person
who originally had them (Posner 1980). That would require us to compare the welfare
gains and losses of different individuals for the goods being transferred. But the source of
the social gain in a contract is not to be found in any such interpersonal comparison, at
least if we restrict ourselves to the economist’s idea of Pareto superiority. Rather, the
social gain in a contract is to be found in intrapersonal comparisons only, namely, in the
idea that no one person (given contractual compensation) is worse off in terms of his or
her own welfare or satisfaction after the exchange than that person was before it, and at
least one person (willing to pay the compensation) is better off. These intrapersonal
comparisons are quite consistent with the possibility that, if an interpersonal comparison
were made, one of the goods or services exchanged was actually going from someone
who valued it more highly in welfare terms to someone who valued it less highly. To see
5
the point vividly, imagine someone who is both cold and hungry, and who, in exchange
for some much needed food, gives up her warm jacket to another person who is neither
cold nor hungry. One can easily appreciate that this might be a Pareto superior exchange
in which each party is better off, but that, nevertheless, the jacket was more highly valued
in an interpersonal sense by the person who was cold. This is a significant point for
understanding the value of Kaldor-Hicks efficiency and we will have reason to reconsider
it below.
Of course, contracts are not the only way in which Pareto superior re-allocations
can be made, and Pareto efficiency achieved. In theory some legal authority could simply
mandate some new state of affairs in which some one was better off and no one was
worse off, perhaps because financial compensation was required to be paid to the latter.
Thus, the voluntarism that is present in market contracting, and which might be attractive
to a libertarian concerned about rights, is not strictly necessary for the achievement of
Pareto superior re-allocations of goods. But economists do tend to prefer the use of
contracts where possible, since the information needed to assess how a proposed reallocation might affect the welfare or satisfaction of individuals is often difficult for third
parties to obtain. On the other hand, if all individuals agree to some re-allocation, the
general view amongst economists is that this must be because, in their own best
judgement, each individual considers himself or herself better off after the re-allocation
than before it or, at least, not any worse off.
Nevertheless, it will sometimes be difficult for the parties to contract so as to
achieve the Pareto superior state of affairs. This will typically be because there are
transactions costs, namely, information costs, negotiation costs, and enforcement costs in
6
the contracting process. Suppose, for example, that downstream users have a right to
receive clean stream water from upstream users, but that an upstream polluting factory
would rather, because it is cheaper than cleaning up the stream, pay the downstream users
to use an alternative supply of bottled water. Suppose too that the downstream users
would be quite happy with this solution to their clean water needs. Nevertheless, it may
be difficult for the firm to identify all the different downstream users (an information
cost), to bargain with each and every one (a negotiation cost), and to strike a bargain that
has terms that are both easy for the parties to observe for compliance and to verify to a
court (the costs of enforcement). Thus, the Pareto superior bottled water solution might
not arise by contract. In such circumstances it might be better to have the same
reallocation of resources achieved in some other way that avoids the costs of contracting,
perhaps by regulation, or by a court injunction providing for some new entitlement. (Of
course, these other methods of resource allocation will have their own sorts of
transactions costs to be considered.) A court could mandate that the downstream users
use bottled water, for example, or, more likely, simply not enjoin the firm from polluting
the stream, leaving the downstream users to buy bottled water on their own as the next
best alternative to having it paid for by the firm. In this way, the court appears to mimic
the market outcome (Coleman 1988). The same allocation of resources that the parties
would have achieved by way of contract (had the costs of contracting not been
prohibitive), namely, the use of bottled water by the downstream users, is achieved by
way of a court decision. The court simply gives to the polluting firm the right to pollute, a
right that the firm would have bought from the downstream users (by financially
7
compensating the downstream users for the costs of bottled water) had the costs of
contracting been less.
Of course, there is a crucial difference between achieving this outcome by way of
a mandatory order and achieving it by contractual agreement with the downstream users.
One has only to ask the downstream users! In the contract, the polluting firm has to make
a payment to the downstream users to use the bottled water, something it was prepared to
do and they were prepared to accept. Thus, in the contractual solution, there is a truly
Pareto superior move. No one is worse off, and someone is better off. However, in the
court’s solution, the downstream users are given a dirty stream and left to pay for the
substitute bottled water on their own. They are clearly worse off than they were before
the court’s mandatory order.
The economist often describes the mandated solution as involving Kaldor-Hicks
efficiency, or potential Pareto superiority rather than actual Pareto superiority (Posner
1980). After all, but for the costs of actual contracting, the parties could have voluntarily
bargained to the result where bottled water was used and the stream was allowed to
become dirty. Under the court’s mandated solution, the argument goes, we have achieved
the same result in the allocation of resources, or the same amount of wealth. The only
difference is that the downstream users have not actually been financially compensated
by the polluter for the costs of substituting bottled water for stream water, although they
could have been. This points to the difference between potential Pareto, or Kaldor-Hicks,
superiority and actual Pareto superiority.
8
II.
Wealth versus Welfare
Now, it is tempting for the economist to dismiss this difference between an actual
Pareto superior gain and a potential Pareto superior gain as merely involving a question
of distribution. After all, given the same resulting allocation of resources, it seems that
there must be the same total social gain. The difference, it might be thought, is only that,
when the new allocation is mandated, and the downstream users are not contractually
compensated, they do not share in the social gain in the way that they would if there was
a contract.
However, it is wrong to think that this is the only difference between a Pareto
superior reallocation and a Kaldor-Hicks or potential Pareto superior one. In fact, if the
courts only mimic the market in a Kaldor-Hicks superior way, and do not insist on the
actual Pareto superiority that is achieved when financial compensation is paid, it is
possible that there may be no social gain in welfare at all. This must surely make us
wonder about what is achieved by way of Kaldor-Hicks efficiency.
To see the problem, one needs to remember the earlier point that a Pareto superior
contractual exchange is quite consistent with there being a re-allocation of some good
from a person who values the good more highly to a person who values the good less
highly in terms of human welfare. This point is often obscured when the economist
speaks about the Kaldor-Hicks re-allocation as generating sufficient gains for the
beneficiaries of the re-allocation such that, hypothetically, the beneficiaries could (by
contract) compensate the losers from the re-allocation so as to render them indifferent to
it and still have some gains left over for themselves. While not, strictly speaking,
9
inaccurate, this way of characterizing the social gain suggests that the welfare gain to the
beneficiaries of securing the good is larger than the welfare loss to the losers of losing it,
an interpersonal comparison. After all, the argument goes, the beneficiaries could
compensate the losers for their losses and still have some of their own gains left over.
But, as already argued, the source of the social gain in a Pareto superior exchange is to be
found in intrapersonal, not interpersonal, comparisons of welfare. It might be that the
person giving up the good valued that good more highly than did the person receiving it,
but that this social loss was more than made up for in the contract by the fact that the
person giving the good up nevertheless valued the financial compensation that she was to
receive in the contract (or, more specifically, what that money would buy for her
elsewhere) even more than she did the good transferred. However, in a Kaldor-Hicks
superior re-allocation, where only the good or entitlement is exchanged, but financial
compensation is not actually provided, we have lost the assurance that any possible loss
in the transfer of the good to someone who values it less highly has been properly offset
for that person by the welfare gain she receives in having money to spend on other goods
she prefers. Thus, it is possible that all we have accomplished is the transfer of a good
from someone who values it more highly to someone who values it less highly and,
therefore, a loss of total welfare overall.
The prospect of a loss of total utility overall within a given re-allocation raises an
even more serious question. Could not a series (or a system) of such Kaldor-Hicks reallocations lead to an outcome in which everyone is worse off than when they started, that
is, to a Pareto inferior state of affairs? If this could be shown to be possible and, further,
possible given some plausible assumptions, then we should surely begin to wonder if the
10
notion of Kaldor-Hicks efficiency has not taken us too far from what was originally so
attractive about Pareto superiority and Pareto optimality.
III.
Kaldor-Hicks Efficiency and Pareto Inferiority
The easiest way to see the possibilities is by recognizing, first, that competitive
market contracting is a kind of Kaldor-Hicks superior allocation. Competitive contracting
is contracting between two parties that does affect a third party. Specifically, the third
party is less well off than she was before the contract because she has lost the business
that she previously had with one of the parties. In economic terminology, she is the
victim of a pecuniary externality, the stuff of market competition. However, economists
are not inclined to worry about this sort of externality (as opposed to a technological
externality like pollution) because the external effect is priced in the market and the good
is allocated according to willingness to pay (Posner 1980). The loser in the competition
lost the benefit of the contract because she had a lesser willingness to pay (or, perhaps, a
lesser willingness to reduce prices) than the competitor who secured the contract. Again,
this will be because she had better uses for her money elsewhere, another sort of
intrapersonal comparison. It is not because, in some interpersonal sense, the good
available in the contract was more valuable to her competitor than it was to her. The
competition is Kaldor-Hicks efficient because the winner in the competition was
obviously prepared to pay more for the contract than the loser was and, therefore, more
than what the loser would have needed to be compensated for choosing her next best
alternative.
11
Let us denote the state of affairs before this new contract (contract #1) as z. This
is the state of affairs that prevailed before our competitive loser, call her C, lost her prior
business with A to her new competitor B. Denote the new post-contractual state of affairs
as y. Obviously, both A and B, the contractors, prefer y to z; C, the competitive loser,
prefers z to y. Or, equivalently, they are better off in terms of their preferences and she is
worse off in terms of hers. Thus, while this process of contracting does result in a KaldorHicks superior re-allocation, it does not bring about one that is Pareto superior.
Now, once in state of affairs y, suppose, in a fully analogous way, that individuals
A and C sign their own Kaldor-Hicks contract (contract #2) to their mutual advantage,
bringing about state of affairs x, which is disadvantageous to B. Finally, imagine further
that in state of affairs x, B and C sign a Kaldor-Hicks efficient contract (contract #3) to
their mutual advantage, which is to the disadvantage of A, producing state of affairs w as
the final social outcome. Thus, in a series of three different Kaldor-Hicks efficient
contracts, we have moved through the different states of affairs from z to y to x to w. The
impact of these three different contracts on the preferences of the three individuals,
together with what we can surmise about these preferences given an assumption that
individual preferences are transitive, is summarized in Table 1.
Table 1
Preferences as Revealed by Three Kaldor-Hicks Efficient Contracts
Assume three Kaldor-Hicks efficient contracts from state of affairs z to y, from y to x, and
from x to w:
Contract #1
Contract #2
Contract #3
By transitivity
Individual A
y preferred to z x preferred to y x preferred to w x preferred to z
Individual B
y preferred to z y preferred to x w preferred to x
Individual C
z preferred to y x preferred to y w preferred to x w preferred to y
12
The important thing to note about these preferences is that for each and every
individual the preference ranking between state of affairs z, where the whole contractual
sequence started, and state of affairs w, where it finished, is so far undetermined. We are
now ready to pose the question: Could it be that this sequence of Kaldor-Hicks efficient
contracts has only resulted in a Pareto-inferior state of affairs, that is, one in which each
individual is worse off than when the sequence started, and over which all individuals
prefer the original state of affairs z?
To see that such a result is not inconsistent with the preferences implied by the
three Kaldor-Hicks efficient contracts, it is enough if we can explicitly introduce this
possibility (that is, the possibility that, for all individuals, z is preferred to w) into the
preferences shown in Table 1 without producing a contradiction (Chapman 1994). Table
2 shows the result, a result that allows for a complete and fully transitive determination of
each individual’s preference ordering over all four of the possible states of affairs (now
ordered by preference for each individual from top to bottom in one of the three
columns).
13
Table 2
Kaldor-Hicks Contracting and Pareto Inferior Outcomes
Assume the preferences in Table 1 and, additionally, that all individuals prefer
state of affairs z to state of affairs w:
Individual A
Individual B
Individual C
x
y
z
y
z
w
z
w
x
w
x
y
Thus, given these individual preference orderings over the four states of affairs,
the three Kaldor Hicks efficient contracts can take these individuals from z to y, from y to
x, and from x to w, and generate a Pareto inferior outcome, or an outcome (w) in which
every individual is worse off than when he or she started (in outcome z).1
Now it might be thought that these must be particularly odd preference orderings,
and that this sort of perverse Kaldor-Hicks sequence should not generally be expected. In
fact, there is something special about the profile of individual preferences shown in Table
2, although in the context in which this preference profile is usually discussed it is not
thought to be particularly odd. The preferences that these three individuals have over the
alternative states of affairs w, x, y, and z, are the sort of preferences that generate the
majority voting paradox, a problem commonly discussed in public and social choice
theory (Mueller 1989). One can see the problem if one imagines that, instead of the three
individuals choosing by contract to move between the various states of affairs, they do so
14
by majority vote. By inspection of Table 2, we can see that a majority of the voters,
considering the various states of affairs in pairs (so as to ensure a majority), would
choose y over z, x over y, w over x, and (if they could reconsider a previously rejected
alternative) z over w (unanimously). Thus, every alternative has another alternative that is
preferred to it by a majority, and it would be impossible for a majority of these
individuals to choose anything but a minority preferred alternative. That is the paradox of
majority voting. Of course, if the group adopted the rule that they would not reconsider or
revisit an alternative that had previously been rejected by a majority (perhaps doing so
because they did not want to cycle endlessly from one alternative to another that was
majority preferred to it), then they might well end up choosing a Pareto inferior outcome.
Indeed, that is exactly what they would do if they followed the same sequence as our
earlier Kaldor-Hicks contractors, going by majority vote from z to y, from y to x, and
from x to w. In state of affairs w, everyone is worse off than they were in state of affairs z
where the sequence began.
Thus, the Table 2 preference profile, which has been shown here to be
problematic for Kaldor-Hicks superior re-allocations (in that a series of such contractual
re-allocations can produce a Pareto inferior outcome), is the same profile that is
commonly discussed, and accepted, as problematic for majority voting. Thus, the
literature in public and social choice theory that has analyzed this preference profile in
some detail is relevant not only to voting mechanisms, where decisive majorities move
groups of individuals between states of affairs (often at the expense of minorities), but
also to Kaldor-Hicks efficient contracting, where the same moves are accomplished by
decisive pairs of contracting partners (often at the expense of competitors). (This should
15
not be surprising, of course, given the general nature of the impossibility results in social
choice theory; see Arrow (1963).) It is worth considering, therefore, what else this
literature can tell us about the value of pursuing Kaldor-Hicks efficiency.
IV.
Value-Restricted Preferences and the Need for Social Consensus
The first insight that this literature provides is that the Table 2 preference profile
is problematic because of the pattern of preferences that it reveals across all the
individuals taken together rather than because of the particular preference ordering that
any one individual might have. Specifically, the difficulty is that the preference profile of
the three individuals in Table 2 fails to manifest the very particular kind of social
consensus that is required if social choice problems, like the Kaldor-Hicks and majority
voting paradoxes, are generally to be avoided.
Public choice economists, when they discuss this problem in a voting context,
emphasize the idea that the problem in Table 2 is that the individual preferences there are
not single peaked (Mueller 1989). When preferences are single peaked, this means that
all the individuals agree that, for every set of three alternatives, (1) there is a single
attribute with respect to which all three alternatives for choice can be decisively ranked,
and (2) one of the three alternatives has an intermediate amount of that decisive attribute.
If both of these conditions are satisfied, then the intermediately placed alternative will be
either best (for those moderates who most prefer an intermediate amount of the decisive
attribute) or in between (for those extremists who most prefer either a lot, or a little, of
the decisive attribute). But the intermediately placed alternative will never be a worst
16
alternative of the three for any of the individuals, that is, one with two more preferred
alternatives (or two peaks) on either side. (If the intermediate alternative was a worst
alternative for some individual, such that the individual preferred either extreme to it,
then that individual would be revealing either that the dimension was not decisive for him
after all, or that this alternative was not, in his view, the intermediate one on that decisive
dimension; how else could one explain him ranking both of the extreme alternatives
ahead of the intermediate one?) For this reason, the single peaked preference requirement
is also sometimes called not-worst value restriction. If this form of value restriction is
satisfied, the arguments show, paradoxes of social choice, like the majority voting
paradox or the Kaldor-Hicks contracting paradox, cannot occur.
As it happens, and as might be suggested by a process of reasoning analogous to
that which generates this result on not-worst value restriction, it can also be proved that
these paradoxes will not occur if all the individuals agree, for every set of three
alternatives, that some one alternative of the three is either not-best, or not-between the
other two. In other words, for the paradoxes to occur, it must be that, for at least one set
of three alternatives, each of the three alternatives appears as best, as between, and as
worst in the preference ordering of at least one of the individuals. When this occurs, it is
sometimes said that the preference orderings over those three alternatives form a Latin
square (Sugden 1981). In Table 2, one can see that there is a Latin square for the set of
three alternatives (x, y, z).
Put in this conjunctive way, it might seem that the paradoxes will be of theoretical
interest only, since they appear to depend too much on a very specific and unlikely sort of
preference profile. However, as the social choice literature shows, this is mere wishful
17
thinking. In multidimensional choice problems, if individuals’ preference orderings
satisfy some very plausible requirements (for example, that individuals prefer alternatives
according to how “close” they are to their ideal point), then if the space of available
alternatives is sufficiently “rich” (or dense), it is almost certain that a problematic Latin
square will appear somewhere over some set of three alternatives (McKelvey 1976). Only
if the individuals’ ideal points all lie on the same line through the multi-dimensional
space will the problem be avoided. However, this effectively reduces the
multidimensional problem to a single dimension or attribute (represented by the equation
of that line), and avoids the social choice problems herein described by reverting to a
kind of not-worst value (or single peaked preference) restriction on that dimension. It
seems much more likely that in a multi-dimensional choice problem (and how many
problems can be reduced to one dimension?) the ideal points of the different individuals
will not lie along the same one line and, therefore, the problematic preference profile, or
Latin square, will be present.
The problems inherent to a multi-dimensional choice problem and, more
particularly, the idea that such a choice problem may generate the special diversity of
views that we see in a Latin square, might be obscured for an economist who equates
Kaldor-Hicks efficiency with the idea of wealth maximization. After all, the goal of
wealth maximization appears to be uni-dimensional. At most it would seem that there
would only be disagreements about a particular wealth maximizing move which imposed
some short term loss on some particular party. But what individual would not want to
maximize wealth over the long run, or over a sequence of such wealth enhancing reallocations, especially if there was a real prospect of sharing in this overall long run gain
18
in social wealth? However, what Table 2 shows is that wealth enhancing re-allocations,
as measured by the greater willingness to pay of the beneficiaries as compared to those
burdened by the re-allocation, are quite consistent with the diversity of views that is
characteristic of, and problematic for, multi-dimensional social choice. Thus, wealth need
not be some one thing that all individuals want more of. Rather, it can be a diversity of
different things, all somewhat misleadingly grouped under this singular word “wealth”.
In the final analysis, what Table 2 shows that it is possible that a series of Kaldor-Hicks
or wealth enhancing re-allocations can add up to an outcome in which we are all worse
off.
This possibility suggests that the economic theorist of legal decision-making
should be interested in finding ways to prevent legal decisions being responsive to the
particular profile of preferences that are represented in Table 2 even if this means
denying some Kaldor-Hicks efficient re-allocations. To the extent that a concern for the
existence of a restricted preference profile is akin to a concern for whether or not shared
values are informing our legally privileged choices, then we are asking the economist to
show more interest in the relationship that should exist between law and social morality.
This, of course, has been the stuff of longstanding debate in legal theory (Hart 1963;
Devlin 1965), and the thought that economic analysis might make some sort of
contribution to that debate should not be entirely unwelcome (Chapman 1979).
From the analysis presented here, we can see that one important contribution is
already at hand. With the help of social choice theory, economics can tell us precisely
what sort of social consensus or social morality is required to exist if society is to
consistently achieve its goal of enhancing the welfare (and not just the wealth) of its
19
members. Specifically, for every set of three alternatives for choice, every individual
should agree that one of the alternatives is either not worst, or not best, or not-between
the other two.
However, we have seen that it is unlikely that this sort of social consensus will
arise as a matter of brute fact, and it would be interesting, therefore, if this analysis could
suggest some institutional requirements for the legal reinforcement of the required social
morality. For this the public choice literature should be useful. For example, it has been
shown that by restricting variations between alternatives for choice to one dimension at a
time, majority voting can avoid some of the difficulties that it would otherwise confront
in a multidimensional choice problem if changes on all dimensions can be proposed
simultaneously (Strom 1990). (Under such a method of choice the alternative finally
chosen is that one that is at the intersection of all the different median voters’ most
preferred amounts on each dimension.)
Further, in the context of corporate law, it has been argued that if managers’ and
directors’ fiduciary obligations are interpreted as being owed to stakeholders more
broadly, and not just to shareholders alone, then certain Kaldor-Hicks contractual
coalitions, which might otherwise threaten the corporation with Pareto inferior outcomes,
are less likely to form (Chapman 1993). The problem is particularly acute in so-called
“end games”, where shareholders seek the cooperation of the firm’s managers in the sale
of their shares to an opportunistic acquirer. This last argument, of course, focuses the
analysis more on the benefits that a firm might gain from recognizing its own corporate
culture, and is less about the advantage that society might gain from reinforcing a
common social morality. But the lesson for these different sorts of collectivity is
20
essentially the same. Without some restrictions on the opportunities that are made
available to decisive coalitions, be they voting majorities or contracting parties in pursuit
of Kaldor-Hicks efficiency, there is a real danger that the cumulative effect of otherwise
uncoordinated decisions will be a Pareto-inferior state of affairs, that is, one in which
everyone is worse off than they otherwise might be.
V.
Questioning the Very Idea of Pareto-Superiority
To this point the argument has essentially been that legal rules designed to
advance Kaldor-Hicks efficiency or, equivalently, the maximization of wealth, will only
provide the illusion of advancing human welfare unless certain restrictions on individual
preferences are also accounted for in the law. While this argument might mean that our
enthusiasm for Kaldor-Hicks efficiency and wealth maximization has to be somewhat
qualified, we can still be grateful that the tools of economic analysis have been useful in
informing, in a very precise way, an age-old debate about the relevance of social
consensus and morality for law. Given a sufficiently rich diversity of individual
preferences, the rights to form certain decisive coalitions, be they majority voting
coalitions or Kaldor-Hicks efficient contractual coalitions, may have to be restricted if
losses in human welfare, and Pareto-inferior states of affairs, are to be avoided.
This last way of putting the point may suggest that there is a normative trade-off
between, on the one hand, respecting certain individual rights and freedoms and, on the
other hand, securing certain gains in overall human welfare, even (Pareto superior) gains
in which everyone has some sort of share. That there is this tradeoff is easy enough to
21
demonstrate. The following example is a variation on one offered by Amartya Sen (Sen
1970). It illustrates, in the context of certain preferences, the possibility of an
incompatibility between a very minimal conception of individual rights and the
satisfaction of the principle of Pareto superiority.
Suppose that there are multiple copies of some controversial book at some public
library. The book is controversial because of its content. Some would like to see the book
banned from the shelves or, at least, kept away from impressionable readers. Others, who
think the book is a truly great piece of literature, would like to see it much more widely
read. (If it helps to have an example of such a controversy, one might think of the book
Lolita by Vladimir Nabokov; Sen’s original example, Lady Chatterley’s Lover, now
seems somewhat dated.) Everyone has the right to read the book, or not, as he or she
chooses. Consider two individuals, Patrick and Herbert, each of whom is representative
of one side of the controversy. Patrick, who objects to the book, would prefer most that
no one read it (state of affairs n), but, failing that, would prefer that only he read it (state
of affairs p) rather than that the younger, more impressionable, Herbert read it (state of
affairs h). Patrick’s least preferred state of affairs, predictably, is that everyone would
read the book (state of affairs e). Herbert, on the other hand, would prefer most that
everyone would read the book, and considers the state of affairs in which no one reads it
to be worst of all. But, if only one sort of person should go to the library and read the
copies that are there, Herbert thinks that it would be better that people like Patrick should
go. He thinks that they have so much more to learn from reading the book than do more
liberal people like Herbert. Thus, where Patrick ranks the four states of affairs in the
order (from left to right) n, p, h, e, Herbert ranks them in the order e, p, h, n.
22
What would happen if these individuals simply exercised their rights of choice to
read the book or not? Although we are interested, ultimately, in the possibility of a
normative tension between rights and welfare, the strategic aspects of the situation are
easily represented as a prisoner’s dilemma (in Table 3), and with predictable results.
Patrick (and everyone else like him) prefers row 2 to row 1, regardless of what column
Herbert chooses (we say, therefore, that row 2 is dominant for Patrick); thus, he chooses
not to read the book. Herbert (and everyone else like him) prefers column 2 to column 1,
regardless of what row Patrick chooses (column 2 is dominant for Herbert), and so
Herbert chooses to read the book. The result is the intersection of row 2 and column 2, or
outcome h, the state of affairs in which only Herbert (and others like him) read the book.
Note, however, that everyone, including Patrick and Herbert (and everyone like them),
considers this outcome to be inferior to outcome p, the state of affairs in which only
people like Patrick read the book. Thus, the exercise of individual rights by all those
involved has resulted in a Pareto-inferior state of affairs.
Table 3
Individual Rights and Pareto Inferiority
Patrick
1. Reads the book
2. Does not read the book
Herbert
1. Does not read the book
2. Reads the book
p
e
n
h
23
Some legal authority could, of course, mandate that only people like Patrick read
the book and, therefore, make everyone better off in terms of their own preferences.
However, there are obvious practical difficulties. Every Patrick has an incentive to
misrepresent himself as a Herbert so that he is not required to read the book. And every
Herbert has an analogous incentive to misrepresent himself as a Patrick so that he is
permitted to read the book. Each of these non-cooperative strategies makes sense for each
of the individuals as it tends to move the individual either towards his most preferred
outcome (if no other individual is adopting this non-cooperative strategy) or away from
his least preferred outcome (if all the other individuals are likewise adopting the noncooperative strategy). In other words, the prisoner’s dilemma structure of the problem
continues to undermine any such mandatory solution. Similarly, any solution to this
problem based on contract (Kaplow and Shavell 2001), where the various Herberts and
Patricks have the additional right to contract with one another to achieve the Pareto
superior state of affairs (where only the Patricks, and not the Herberts, read the book), is
going to be subject to this difficulty. After all, if there is a good to be achieved here in
having Patrick and others like him read the book, then it is a public good (that is, one for
which the benefits of consumption cannot be made available only to those who
contributed to the costs of its provision) and, therefore, one that is susceptible in the usual
way to individuals free riding on the cooperative behaviour (contractual commitments) of
others. Of course, if everyone chooses to free ride, then the public good is not provided
and everyone is worse off, a Pareto inferior state of affairs.
But the real point of the example is not to spell out a familiar strategic problem,
but rather to suggest that the principle of Pareto superiority may not be so sacred after all.
24
If we believe that there is some normative value in allowing people to choose, as a matter
of individual right, to read the books they want, then the possibility that everyone might
be worse off than they would be if they did not have such rights might well be an
acceptable outcome. Of course, this would mean that the normative foundation for such
legal rights is grounded in something other than welfare. However, this may not be much
of a surprise to anyone but the most avid and single-minded proponent of economic
efficiency as a normative foundation for law.
25
BIBLIOGRAPHY
Arrow, Kenneth J. (1963), Social Choice and Individual Values, New Haven and
London: Yale university Press
Chapman, Bruce (1979), ‘Law, morality, and the logic of choice: an economist’s view’,
University of Toronto Law Journal, 29 (2), 114-137.
------------------- (1993), ‘Trust, economic rationality, and the corporate fiduciary
obligation’, University of Toronto Law Journal, 43 (3), 547-588.
-------------------- (1994), ‘The rational and the reasonable: social choice theory and
adjudication’, University of Chicago Law Review, 61 (1), 41-122.
Coleman, Jules L. (1988), Markets, Morals and the Law, Cambridge and New York:
Cambridge University Press.
Devlin, Patrick (1965), The Enforcement of Morals, Oxford and New York: Oxford
University Press.
Hart, H.L.A. (1963), Law, Liberty and Morality, Oxford and New York: Oxford
University Press.
26
Kaplow, L. and Shavell, S. (2001), ‘Fairness versus welfare’, Harvard Law Review 114,
961-
.
Landes, William A. and Posner, Richard A. (1987), The Economic Structure of Tort Law,
Cambridge, Mass. and London: Harvard University Press.
McKelvey, R.D. (1976), ‘Intransitivities in multidimensional voting models and some
implications for agenda control’, Journal of Economic Theory, 12, 472 - 482.
Mueller, Dennis C. (1989), Public Choice II, Cambridge and New York: Cambridge
University Press.
Peltzman, Sam (1975), ‘The effects of automobile safety regulation’, Journal of Political
Economy, 83, 677 -
.
Posner, Richard A. (1980), ‘The ethical and political basis of the efficiency norm in
common law adjudication’, Hofstra Law Review, 8, 487 – 507.
Sen, Amartya (1970), ‘The impossibility of a Paretian liberal’, Journal of Political
Economy, 78, 152 – 157.
Strom, Gerald S. (1990), The Logic of Law Making, Baltimore and London: Johns
Hopkins Press.
27
Sugden, Robert (1981), The Political Economy of Public Choice, Oxford: Martin
Robertson & Co.
Trebilcock, Michael J. (1997), ‘An introduction to law and economics’, Monash
University Law Review, 23 (1), 123 – 158.
Zerbe, Richard O. (2001), Economic Efficiency in Law and Economics, Cheltenham, UK
and Northhamptom, Mass.: Edward Elgar.
1
Two points are worth emphasizing at this point to avoid confusion. First, the individuals might not be
content to stop at outcome w if it makes them all worse off. They might prefer to return to outcome z. But
this is only to say that the problem might manifest itself as a cycle rather than as a choice sequence leading
to (and ending in) a Pareto-inferior state of affairs. This alternative scenario is familiar enough to those who
study the fully analogous problem in voting theory, namely, the majority voting paradox. On this last point,
see the paragraph to follow in the text.
Second, it might be thought impossible to distribute wealth, as measured by willingness to pay,
over the three transactions (from z to y, from y to x, and from x to w), so that everyone is made worse off
than when they started. However, the possibility is real, and it turns on taking the “ability to pay” constraint
seriously. Having “paid” for some beneficial transactions, one runs out of the wealth needed to resist the
(most) harmful transaction when it appears in the transactional sequence. Of course, one does not actually
pay in Kaldor-Hicks efficient transactions, but if willingness to pay is going to mean anything, one has to
take the idea of a budget constraint seriously. So, in the table below (which reproduces Table 2 from the
text with the following additions), let the willingness to pay of the different individuals for the different
transactions be represented by the letters in bold immediately to the right of that individual’s preference
ordering. (For ease of exposition, and without loss of generality, the willingness to pay of each individual
for z over w has been assumed to be the same, namely, c; c > 0 if z is Pareto-superior to w.)
Individual A
x
Individual B
y
a
y
b
c
w
c
w
c
w
z
d
z
z
Individual C
f
x
e
x
g
y
Thus, for example, individual A and individual B would pay b and d respectively for the transaction from
state of affairs z to y, and individual C would pay all she can, or c + f + g, to avoid that transaction (the
worse possible for her in this scenario). So, if this transaction goes through on Kaldor-Hicks grounds, it
must be that:
28
(1)
b+d > c+ f + g
Analogously, the Kaldor-Hicks approved transaction from y to x implies that the willingness to pay of
individual A and individual C for this transaction, or a and g respectively, is larger in aggregate than what
individual B is willing (and able) to pay to avoid this transaction, taking into account (or subtracting) that
individual B has already spent d in the earlier z to y transaction. In symbols, we have:
(2)
a+ g > d+ c+ e-d
Finally, the Kaldor-Hicks approved transaction from x to w implies analogously (taking into account that
individual A has already spent a and b in the earlier transactions) that:
(3)
e + f > a + b + c – (a + b)
Gathering terms on each side of these inequalities implies that:
(4)
b + d + a > 3c
Now, obviously, if the final outcome state of affairs w is Pareto-inferior to the original state of affairs z,
then the term on the right hand side is positive (since c > 0). Thus, a Kaldor-Hicks approved sequence of
transactions can lead to a Pareto-inferior state if and only if the terms on the left of the inequality are
positive and larger than the terms on the right. But there is no reason to think that this is a problematic
requirement. Imagine only that c, while positive, is small. Then almost any willingness to pay by
individuals A and B for the move from z to y, or any amount that A might pay for the move from y to x,
would be sufficient for the result. Of course, this does suggest that A and B are slightly myopic (in that the
choice sequence they undertake will end up making them worse off), but that is consistent with the usual
demonstration of a majority voting paradox that leads a politically decisive majority to choose a Paretoinferior outcome; on this, see again the next paragraph in the text.